On Tue, Apr 6, 2010 at 1:27 PM, Laurent Pinchart <laurent.pinchart@xxxxxxxxxxxxxxxx> wrote: > Hi Andy, > > On Tuesday 06 April 2010 13:06:18 Andy Walls wrote: >> On Tue, 2010-04-06 at 08:37 +0200, Hans Verkuil wrote: > > [snip] > >> > Again, I still don't know whether we should do this. It is dangerously >> > seductive because it would be so trivial to implement. >> >> It's like watching ships run aground on a shallow sandbar that all the >> locals know about. The waters off of 'Point /sys' are full of usability >> shipwrecks. I don't know if it's some siren's song, the lack of a light >> house, or just strange currents that deceive even seasoned >> navigators.... >> >> Let the user run 'v4l2-ctl -d /dev/videoN -L' to learn about the control >> metatdata. It's not as easy as typing 'cat', but the user base using >> sysfs in an interactive shell or shell script should also know how to >> use v4l2-ctl. In embedded systems, the final system deployment should >> not need the control metadata available from sysfs in a command shell >> anyway. > > I fully agree with this. If we push the idea one step further, why do we need > to expose controls in sysfs at all ? > how about security permissions? while you can easily change the permission levels for nodes in /dev you can't do this so easily with sysfs entries. I don't really think this is needed at all some applications will start to use ioctl some other apps might go for sysfs.. this makes the API a little bit whacko Markus -- To unsubscribe from this list: send the line "unsubscribe linux-media" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
